05 Operations
Old board retired + vault connections wired (2026-10-20, Young ruling)
Old board retired + vault connections wired (2026-10-20, Young ruling)
What moved
solos-v3board REMOVED on both hosts (soft-archived, recoverable): LAB to_archived/solos-v3-1792495130, laptop to_archived/solos-v3-1792495138. The old V3 build board no longer claims space in the live kernel. Content survives in the archive tombstones (45 archived + 3 blocked rows recorded, recoverable if the ledger ever needs the tails).- Vault kanban projections wired BOTH units, per ruling:
- LAB agents (sol/min/mira/nova/echo) ->
Docs/Kanban/- three .kanban projections regenerated live into the vault and committed:docker-vhost.kanban,research.kanban,daily-operation.kanban(c966cc6). Sync script deployed durably at/home/lab/.hermes/profiles/min/scripts/(sync_board_projection.py + per-boardproj_<board>.sh), NOT in scratch (prunable). - PULSE agents (eva/eve/tam) ->
Pulse/Kanban/-daily-report-brief.kanban,research.kanban,podcast.kanbanregenerated on the laptop and pushed into the vault (485ac63). Script deployed atprofiles/tam/scripts/on laptop (HERMES_BIN pinned absolute, per the sync-kanban-projection gotcha: plain cron PATH misseshermes).
- LAB agents (sol/min/mira/nova/echo) ->
- 30-minute no-agent crons registered for all six boards (LAB 3 + laptop 3) - mode deliver=local per the standing fleet-cron rule (quiet unless failed); the cron registry now carries: LAB {docker-vhost, research, daily-operation} + laptop {daily-report-brief, research, podcast}.
- Migration old-to-new (Young-s last row): the old solos-v3 board-s live work
(3 ready / 6 running cards) already re-seeded EARLIER TONIGHT onto the new boards
in their correct lanes. The one-time
SOLOS-V3.mdvault projection was removed and replaced by per-board .kanban files. No task was dropped in the move - each had a correct new home before the old board was retired.
Verified ground truth
- LAB kanban boards: daily-operation ready=5, docker-vhost ready=4 (running live on their cards), research ready=1, default empty.
- Laptop kanban boards: daily-report-brief ready=2, research ready=2, podcast ready=2
- pulse/default archives.
- Projections on disk with sha256 markers + "kernel is truth" preamble in each file.
Honest limits
- First cron fires land within 30 min (next run timestamps registered); a projected file mid-build can be up to 30 min behind the kernel by design (the file states it) - the heartbeat polling lane is the freshest view, the vault file is the human-readable slice.
- The LAB dispatcher loop per board awaits Min-s go (cards exist + cron projections exist; worker spawn wiring is the hermes-native-fleet-system blueprint, next beat).
🎯 Tam Read: the old V3 board is retired on both hosts and the vault now carries all six live boards as read-only projections - 3 into Docs/Kanban for the Brain agents, 3 into Pulse/Kanban for the Pulse lane, each refreshed every 30 minutes by no-agent crons with the kernel as the single truth - the fleet tasks all migrated without a single card dropped.
Kanban on Telegram (2026-10-20) — enabled with the correct access gate
The ledger was blind, and it was writing into a dead tree
Young's question, 2026-10-21: "can't be right everything is zero?" — about the daily spend-ledger-daily cron reporting total=$0.0000 paid=$0.0000 for both LAB and prod, every day, with alert=OK.

